Reply

SIR—We appreciate the concern expressed by Jacobs et al. [1] in their perception of an inaccurate clinical message in our recent article in Clinical Infectious Diseases [2] and their suggestion that MIC interpretative breakpoints should be based on pharmacokinetic and pharmacodynamic (PK/PD) parameters. However, we do not agree that our publication [2] should be used as a vehicle by which Jacobs and colleagues promote their arguments about the use of PK/PD parameters.

In describing a study of the 1999–2000 respiratory season and comparing it with a similar study conducted in the 1998–1999 respiratory season, we chose to analyze our data according to the National Committee for Clinical Laboratory Standards (NCCLS) breakpoints that had been in place at the beginning of each respiratory season [3, 4]. A common problem encountered by persons who conduct and publish antimicrobial surveillance studies is that MIC breakpoints for some organism-antimicrobial combinations change frequently (e.g., those for Streptococcus pneumoniae)—sometimes between the time a manuscript is submitted to a peer-reviewed journal and the time that it is published. To counter this problem for the studies described in our recent article [2], we decided to use the breakpoints that were extant at the time each study was initiated, and we clearly documented this decision [3, 4]. The impact of a subsequent revision to the NCCLS breakpoints [5] that were available to us at the time that the manuscript was written was then discussed in the text. In fact, one of the intentions of the article was to demonstrate the impact that recent changes in the NCCLS MIC interpretative breakpoints have had on the perception of antimicrobial resistance among pneumococci. To suggest, as Jacobs et al. [1] have, that interested readers of Clinical Infectious Diseases do not read the text of published articles is clearly mistaken. We also strongly contest the assertion that we have reported falsely high rates of resistance to amoxicillin-clavulanate and cefuroxime, because all methods and interpretative standards used in the completion and publication of this study were referenced appropriately. It is interesting that Jacobs et al. [1] failed to acknowledge the importance of the MIC range, modal MIC, and MIC₉₀ data that appeared in the same article, as well as the MIC distribution data, against which any set of MIC breakpoint criteria can be applied, including different versions of NCCLS documents—yet Jacobs and colleagues used those very data to demonstrate the impact of application of PK/PD breakpoints.

Antimicrobial prescription decisions made on the basis of PK/PD principles that predict efficacy and bacterial eradication and that prevent the emergence of resistance are important to the success of therapy [6, 7]. We agree that the application of clinical and PK/PD breakpoints in defining breakpoints is a way forward, but, in our study, as in other antimicrobial surveillance studies, NCCLS criteria were used to interpret MIC data. Regardless of whether we concur with the current MIC interpretative breakpoints, the breakpoints should be used as they were agreed upon and published by the appropriate bodies (e.g., the US Food and Drug Administration [FDA] and NCCLS in the United States). The FDA and NCCLS choose breakpoints on the basis of many parameters, including PK/PD data. Ideally, MIC distributions for all relevant antimicrobials would accompany every surveillance study to facilitate the application of revised interpretative criteria and criteria from different countries and to facilitate retrospective comparisons. Monitoring of shifts in MICs can be a very sensitive method for determination of changes in susceptibility that depend on the most prevalent mechanism of resistance (i.e., mutation versus gene exchange) in the organism under study.

Finally, although the need for surveillance is obvious, and although its use in the treatment of individual patients is important, it cannot unequivocally predict outcomes in patients with infections. Treatment regimens for individual patients with suspected or demonstrated infections should be developed after consideration of symptoms, laboratory findings, and relevant medical history and in the context of appropriate local and widespread antimicrobial resistance trends. Frequently, patients have many concurrent factors that can influence the outcome of an infection, only one of which is the in vitro susceptibility of the pathogen to the antimicrobial(s) selected for treatment. For an individual patient, the antimicrobial susceptibility of an infecting pathogen may be no more likely to accurately predict clinical outcome than is a radiograph, even though both may be important for patient management.